Subrat Kumar Swain
CV | Google Scholar | GitHub | LinkedIn
s dot swain at uq dot edu dot au
I am a first-year Joint-PhD scholar at The University of Queensland and Indian Institute of Technology, Delhi, advised by Professorss
Vireshwar Kumar and Dan Kim
My area of research lies in the security of machine learning and machine learning for security. Under that umbrella, I'm currently looking
at semantic preserving adversarial attacks, adversarial robustness, and model stealing/ extraction. I'll also be extending these ideas to
machine learning for security, specifically in intrusion detection systems.
Apart from research, I enjoy dancing; lyrical hip-hop. I do sketch sometimes, mostly abstract/ modern art and sometimes portraits. One of
my paintings was selected for the walls of the Ministry of HRD in a national level art showcase. I also like cooking, reading history,
and looking at maps for hours.
And yes, I prefer tabs over spaces :)
Education
Joint-PhD, Computer Science, Aug, 2021 - present
The University of Queensland and IIT Delhi
Courses:Machine Learning, Computer Vision, Meta Learning, Cyber-physical Systems,
Network and System Security
CGPA: 9.00/ 10
Btech, Computer Science, 2016 - 2020
Core Courses: Soft Computing, Mathematics, DS and Algo, OS, Networking, DBMS, Digital Electronics, Computer
Organisation and Architecture, Theory of Computation, Compiler Design.
CGPA: 9.21/ 10
VSSUT, Sambalpur, India
Work Experience
Product Engineer, September, 2020-August, 2021
Cognizant Technology Solutions, Bengaluru, India
✔ Developed and integrated a Ticket Analysis Solution into Nexa, an AutoML platform, which generates plots for volumetric
& multivariate analysis and arrival patterns of the tickets.
Machine Learning Engineer, May, 2019-August, 2020
Taiyo AI, San Francisco, California
✔ Integrated Bayesian Hyperparameter Optimization method in the existing ML
Utils library thereby reducing the convergence time.
✔ Deployed Timeseries Forecasting models of SP500 Stocks, Macroeconomic and
Forex indicators by leveraging Apache Airflow Pipelines on Cloud Composer of GCP.
Publications and Presentations
Deep learning and wavelet transform integrated approach for short-term solar PV power prediction
Manohar Mishra, Pandit Byomakesha Dash, Janmenjoy Nayak, Bighnaraj Naik, Subrat Kumar Swain
Journal of the International Measurement Confederation (IF: 5.131)
[paper]
SPAT: Semantic-Preserving Adversarial Transformation for Perceptually Similar Adversarial Examples
Subrat Kumar Swain, Vireshwar Kumar, Dan Dongseong Kim, Guangdong Bai
Accepted to European Conference on Artificial Intelligence (ECAI) [CORE A](2023)